My own "scientific" poll

[partypooper]

im going to try and call 100 people tomorrow, i have no work right now because of the rain over here in georgia.

one thing i will change about calling tomorrow is instead of just asking them who they are voting for, ask them do they know who they voting for. if they say "i dont know" i will ask them "who are you all interested in right now."

that way i can set up who people are interested in also

since you are going to spend some time and/or money on this i would like to give you some advice:

take a true random sample. here is one way to do it:

http://www.randomizer.org/lesson4.htm

many samples that feel random are not really random, because we are psychologically drawn to certain names and positions. phonebook is not really random either, because it has clusters of last names that might have similarities in their political attitudes. it would be a shame to spend a whole day on the phone for a non-random sample.

make sure you are neutral and do not lead your subjects in any way.

keep track of numbers where people don't respond and call them again. those people might be different. keep track of how many people you had to reach in order to obtain 100 subjects.

of course, even that will get you an unacceptably large margin of error but it is a start and it might continue to rain...

 

[garrettwombat]

i know its not accurate but i just wanna see if anyone does actually say Ron paul, if i get enough people to say his name than i will feel a lot better about the election.

 

[JohnXSmith]

This is pretty much my exact words:

"Hi, I'm John and I'm doing a poll about the upcoming elections. Can I ask you a few questions?" The response has always been yes, ask away.

"Do you plan on voting in the presidential primary?" If they answer no, I say thank you for your time and say goodbye.

If they answer yes, I asked:
"Who will you vote for?"

Note I don't mention any names of candidates. If they choose a candidate, they have to know the name so they can tell me.

For randomness, I close my eyes when selecting the page of the phonebook and with my eyes still closed put my finger somewhere on the page and call that number. Seems like an easy way to get a random phone number.

John

 

[partypooper]

For randomness, I close my eyes when selecting the page of the phonebook and with my eyes still closed put my finger somewhere on the page and call that number. Seems like an easy way to get a random phone number.

technically speaking, that is not random. the assumptiom of "scientific" polling is that you can calculate the probability that any member of the population will end up in your sample. if you choose your sample in this way you can't calculate that probability. this matters partly because of bias (not likely in this case) but partly because formulae for margin of error depend on certain assumptions about how the sample was constructed. if you violate these assumptions you can't calculate the margin of error.

but i guess this is for forum purposes only. let us know what you got.